W_llm_intent - AI Intent Classification (ollama)

The workflow JSON

Copy or download the full n8n JSON below. Paste it into a new n8n workflow, add your credentials, activate. Full import guide →

{
  "name": "W_LLM_INTENT - AI Intent Classification (Ollama)",
  "active": false,
  "settings": {
    "executionTimeout": 15,
    "saveManualExecutions": true
  },
  "nodes": [
    {
      "parameters": {},
      "id": "llm-trigger-01",
      "name": "IN - From Core",
      "type": "n8n-nodes-base.executeWorkflowTrigger",
      "typeVersion": 1,
      "position": [
        0,
        0
      ]
    },
    {
      "parameters": {
        "language": "javascript",
        "jsCode": "const e = $json;\nconst userText = (e.userText || e.message?.text || '').toString().trim();\nconst locale = (e.l10n?.detectedLocale || e.state?.locale || 'fr');\nconst cartItems = (e.cart?.items || []).map(i => i.name).join(', ');\nconst stage = (e.state?.stage || 'START');\n\nconst systemPrompt = `You are RestoBot, a restaurant ordering assistant for an Algerian restaurant.\nYou classify customer messages into exactly ONE intent.\nValid intents: GREET, VIEW_MENU, ADD_ITEM, REMOVE_ITEM, VIEW_CART, CHECKOUT, DELIVERY, SUPPORT, FAQ, LANG_SWITCH, UNKNOWN.\nRespond ONLY with a JSON object: {\"intent\": \"...\", \"confidence\": 0.0-1.0, \"entities\": {}}\nDo NOT include any other text.`;\n\nconst userPrompt = `Customer message (locale: ${locale}, stage: ${stage}, cart: [${cartItems}]):\n\"${userText.substring(0, 500)}\"`;\n\nreturn [{ json: { ...e, _llm: { systemPrompt, userPrompt, userText } } }];"
      },
      "id": "llm-prepare-02",
      "name": "Prepare LLM Prompt",
      "type": "n8n-nodes-base.code",
      "typeVersion": 2,
      "position": [
        220,
        0
      ]
    },
    {
      "parameters": {
        "method": "POST",
        "url": "={{ $json._strapiConfig?.ollama_api_url || $env.LLM_API_URL || 'http://ollama:11434/api/chat' }}",
        "sendBody": true,
        "specifyBody": "json",
        "jsonBody": "={{ JSON.stringify({ model: $json._strapiConfig?.llm_model || $env.LLM_MODEL || 'llama3.1', messages: [{ role: 'system', content: $json._llm.systemPrompt }, { role: 'user', content: $json._llm.userPrompt }], stream: false, options: { temperature: $json._strapiConfig?.llm_temperature || 0.1, num_predict: $json._strapiConfig?.llm_max_tokens || 100 } }) }}",
        "options": {
          "timeout": 10000,
          "response": {
            "response": {
              "fullResponse": true
            }
          }
        }
      },
      "id": "llm-call-03",
      "name": "Call Ollama API",
      "type": "n8n-nodes-base.httpRequest",
      "typeVersion": 4.2,
      "position": [
        440,
        0
      ],
      "continueOnFail": true
    },
    {
      "parameters": {
        "language": "javascript",
        "jsCode": "const e = $json;\nconst fallback = { intent: 'UNKNOWN', confidence: 0, entities: {}, llm_used: false };\n\ntry {\n  const body = e.body || e.data || e;\n  const content = (body?.message?.content || '').toString().trim();\n  if (!content) return [{ json: { ...e, llmResult: { ...fallback, reason: 'empty_response' } } }];\n\n  // Try to parse JSON from the LLM response\n  const jsonMatch = content.match(/\\{[^}]+\\}/);\n  if (!jsonMatch) return [{ json: { ...e, llmResult: { ...fallback, reason: 'no_json_in_response' } } }];\n\n  const parsed = JSON.parse(jsonMatch[0]);\n  const validIntents = ['GREET','VIEW_MENU','ADD_ITEM','REMOVE_ITEM','VIEW_CART','CHECKOUT','DELIVERY','SUPPORT','FAQ','LANG_SWITCH','UNKNOWN'];\n  const intent = validIntents.includes(parsed.intent) ? parsed.intent : 'UNKNOWN';\n  const confidence = Math.min(1, Math.max(0, Number(parsed.confidence) || 0));\n\n  return [{ json: { ...e, llmResult: { intent, confidence, entities: parsed.entities || {}, llm_used: true } } }];\n} catch (err) {\n  return [{ json: { ...e, llmResult: { ...fallback, reason: 'parse_error', error: err.message } } }];\n}"
      },
      "id": "llm-parse-04",
      "name": "Parse LLM Response",
      "type": "n8n-nodes-base.code",
      "typeVersion": 2,
      "position": [
        660,
        0
      ]
    }
  ],
  "connections": {
    "IN - From Core": {
      "main": [
        [
          {
            "node": "Prepare LLM Prompt",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Prepare LLM Prompt": {
      "main": [
        [
          {
            "node": "Call Ollama API",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Call Ollama API": {
      "main": [
        [
          {
            "node": "Parse LLM Response",
            "type": "main",
            "index": 0
          }
        ]
      ]
    }
  }
}

About this workflow

W_LLM_INTENT - AI Intent Classification (Ollama). Uses executeWorkflowTrigger, httpRequest. Event-driven trigger; 4 nodes.

Source: https://github.com/zerAda/RestaurantAgentAutomation/blob/41a4d42dcd66e57b1e87b4750c0fd5fbf7058f68/workflows/W_LLM_INTENT.json — original creator credit. Request a take-down →

More AI & RAG workflows → · Browse all categories →

Related workflows

Workflows that share integrations, category, or trigger type with this one. All free to copy and import.